The global use of ART has resulted in pronounced clinical advantages with decreased morbidity and mortality. However at the global level, both pretreatment and acquired HIV-1 drug resistance (HIV-DR) are increasing (https://www.who.int/hiv/topics/drugresistance/
protocols /en/). Since the development of ARVs has to a large extent been based on HIV-1B, the knowledge about HIV-DR in non-B subtypes is limited. My thesis discusses some of the existing knowledge gaps in the identification and treatment of HIV-1 drug resistance.
Identification and quantification of drug-resistant viral populations before initiation of ART reduces the risk of developing virological treatment failure and acquired DRMs in patients who have PDR. In HIC, population-based Sanger sequencing has been used for a long time, both before the initiation of first-line ART and at ART failure. More recently, HTS has been developed and has so far been introduced into clinical care in a few laboratories. A major issue concerning HTS is the identification cut-offs of clinically relevant resistance for minor viral variants. Recently, a clinically relevant cut off of 5% for NNRTI resistant strains has been suggested, but there is limited knowledge for other categories of ARVs [51]. In LMIC, HIV-DR diagnostics are seldom performed in the clinical setting, but important HIV-DR surveillance programs exist, e.g., run by WHO [51].
To facilitate the analysis of viral sequences obtained through HTS in patients from both HIC and LMIC, we developed a user-friendly bioinformatics pipeline to bridge the gap between clinicians and bioinformatics expertise (Paper I). The output file from the high-throughput platform can be analyzed by uploading it in the MiDRMpol pipeline to get the final output in an easily readable table format. The overall process takes less than 10 minutes and does not require the help of onsite bioinformatics expertise. MiDRMpol can allow the user to choose a subtype-specific reference sequence (A, B, C, D, 01_AE, 02_AG) while variant calling followed by quantification of DRMs.
Due to the increasing global occurrence of HIV-DR and due to side effects of the available ARVs, it is of importance to develop new ARVs, preferable with a high genetic barrier and less toxicity. EFdA is one among the promising new drugs presently entering clinical phase II and III trials. The high genetic barrier of EFdA is due to its strong interactions with the active site of the RT where DRM are prone to develop. In Paper II, we studied its in vitro potency against diverse HIV-1 subtypes and showed a high potency of EFdA against several HIV-1 RTI resistant strains (NIH panel) derived from different HIV-1 subtypes. We conclude therefore that EFdA has a high inhibition potency, irrespective of the subtype. In addition, we tested EFdA against two of the viruses (SE138.4 and SE141.5) from the Swedish cohort which had the K65R mutation and found a higher efficiency (0.8 and 3.4 fold respectively) than TAF. This finding is in accordance with another study, where hypersusceptibility for EFdA was observed in virus with the K65R mutation in patients, who were failing on first-line TDF ART [52]. Thus, although EFdA is not yet approved by regulatory authorities, our data suggest that it can be a valuable addition to the presently used ARVs, both in HIC and LMIC.
The second-generation INSTIs DTG is presently rolled out on a large scale both in HIC and LMIC. However, most existing knowledge about this drug is derived from studies on
HIV-26
1B. Also, the former view that HIV-DR against DTG develops only with very great severity has shown not to hold through [53]. The knowledge about the two other second-generation INSTIs, CAB and BIC with regard to HIV-DR pattern is also limited. However, recently DRM was shown in some patients failing long-acting CAB and RPV within a clinical trial
[54]. In Paper III, we studied the potency of first and second-generation INSTIs against four HIV-1 subtypes (B, C, 01_AE and 02_AG) using biochemical and cell culture assays. The integration of HIV-DNA into the host genome occurs in two major steps: 3’-end processing followed by strand transfer reaction [55]. Our results from the biochemical assays confirmed that all INSTIs are not as efficient in inhibiting 3’-end processing as compared to strand transfer activity for all subtypes, which is in agreement with previous reports for RAL, EVG
[56], DTG [57] and BIC [58]. Interestingly, a natural polymorphism M50I was found in 7/24 (18%) viruses, which is in line with the reported frequency of 10-25% of INSTI-naïve patients in different subtypes according to Stanford HIV drug resistance database [59]. This polymorphism has been observed in combination with R263K in a patient who subsequently failed treatment with RAL [60]. R263K has been shown to increase resistance to DTG by 2-5 folds and decrease viral replication and strand transfer activity [61]. In viral outgrowth assay, M50I emerges after R263K [58]. Whether the presence of M50I provides a replication advantage to R263K[58] or not[62] is still under debate. The combination of R263K/M50I results in 2.8 fold reduction in BIC-susceptibility, but M50I alone does not have any effect
[58]. In our in vitro experiments in Paper III, M50I had no effect on any of the INSTIs used.
One of the solutions for the problem of patients suboptimal adherence to ART is to develop and implement long-acting drugs to allow flexibility in dosing of at least monthly and perhaps longer. Due to a prolonged intracellular half-life of EFdA active moiety in human blood cells, it is potentially possible for this drug [63, 64]. A further two long-acting injectable antiretroviral compounds, CAB and RPV, have completed clinical trials demonstrating safety, tolerability and prolonged antiretroviral activity. In Paper II and III we showed that EFdA and CAB had potent activity against diverse HIV-1 subtypes. In contrast, we have identified that RPV may not be an optimal choice for HIV-1C [65]. Of these three anti-HIV drugs, which have the potential of being used as long-acting drugs, we feel a concern about the use of RPV where HIV-1C dominates the epidemic. However, an even higher threat against the efficacy of long-acting regimens is the high risk of developing HIV-DR to these drugs if the patient does not turn up for the regular therapeutic injections. This risk is based on the pharmacokinetic properties of the drugs where the elimination of the drugs takes a very long time, during which suboptimal drug concentrations will appear [66].
In Paper IV, we showed that virus with PYQE insertion had increased replication capacity compared to non-PYxE-strains in vitro. PYxE (where x =R/K/Q) insertion in the P6 region of the gag is one among the unique features of HIV-1C, which was reported for the first time by Neogi et al. [22]. Both the PYQE variant of PYxE in our experiments and the PYRE inserted HIV-1C reference virus (pIndieC1) in another study [67] showed higher replication capacity.
The PYxE insertion is prevalent among HIV-1C from Ethiopia and Eritrea compared to South African or Indian HIV-1C strains [22]. Studies have also shown that PYxE insertion is more common in viruses from HIV-1C therapy-failure patients than in HIV-1B viruses from failing patients [22]. The mechanism has been studied by Domselaar et al., [68] where the in silico and in vitro experiments show that the PYxE insertion in gag-p6 of HIV1-C increases its binding to ALIX and enhances the viral fitness. It was observed that the increased fitness
27
related to PYxE could affect the sensitivity against the protease inhibitor lopinavir in the absence of any PI DRM. However, a more detailed analysis of the clinical significance of the PYxE insertion within HIV-1C gag is needed.
In conclusion, quantification of viral variants, mainly minor populations in the pool of HIV quasispecies within a patient, is a promising leap to improve the selection of optimal first-line ART, which is likely to result in fewer ART failures. There is a continous development of methods and analysis tools for the determination of HIV-DR, also with the purpose to reduce the cost of sample processing and analysis. MiDRMpol is one such tool based on our intentions to keep it user-friendly and free of charge, as explained in Paper I. The new ARV EFdA shows promising results in suppressing the viral load in in vitro experiments against diverse subtypes of HIV-1, also in multi-resistant strains, as well as in HIV-2 as demonstrated in other studies. Paper II indicates that EFdA is likely to be as efficient in non-B subtypes as in HIV-1B although this should be evaluated in more extensive clinical studies, including patients infected by diverse HIV-1 subtypes harboring different pattern of DRMs. Similarly, in Paper III the second-generation INSTIs CAB and BIC suppressed diverse viral variants in vitro, and the data give support to their future use in countries dominated by non-B subtypes.
Along with the clinically well-defined DRM in the pol region described in Papers I, II and III, there are also other mutations and insertions that may or may not alter the drug susceptibility, but that also have the potential to increase the viral replication capacity increasing the risk of negative clinical consequences. One such genetic change is the PYxE insertion which is unique to HIV-1C as described in Paper IV. Altogether my thesis and the Papers I-IV show the importance of considering the efficacy of ARVs on all HIV-1 subtypes and also include analysis of viral variants in the minor viral populations. If this is done I believe that the impact of ART on the global HIV epidemic will be even larger than it is been so far.
28